Nociceptive sensory neurons drive interleukin-23-mediated psoriasiform skin inflammation

Abstract

The skin has a dual function as a barrier and a sensory interface between the body and the environment. To protect against invading pathogens, the skin harbours specialized immune cells, including dermal dendritic cells (DDCs) and interleukin (IL)-17-producing γδ T (γδT17) cells, the aberrant activation of which by IL-23 can provoke psoriasis-like inflammation. The skin is also innervated by a meshwork of peripheral nerves consisting of relatively sparse autonomic and abundant sensory fibres. Interactions between the autonomic nervous system and immune cells in lymphoid organs are known to contribute to systemic immunity, but how peripheral nerves regulate cutaneous immune responses remains unclear. We exposed the skin of mice to imiquimod, which induces IL-23-dependent psoriasis-like inflammation. Here we show that a subset of sensory neurons expressing the ion channels TRPV1 and Nav1.8 is essential to drive this inflammatory response. Imaging of intact skin revealed that a large fraction of DDCs, the principal source of IL-23, is in close contact with these nociceptors. Upon selective pharmacological or genetic ablation of nociceptors, DDCs failed to produce IL-23 in imiquimod-exposed skin. Consequently, the local production of IL-23-dependent inflammatory cytokines by dermal γδT17 cells and the subsequent recruitment of inflammatory cells to the skin were markedly reduced. Intradermal injection of IL-23 bypassed the requirement for nociceptor communication with DDCs and restored the inflammatory response. These findings indicate that TRPV1(+)Nav1.8(+) nociceptors, by interacting with DDCs, regulate the IL-23/IL-17 pathway and control cutaneous immune responses.

Conflict of interest statement

No authors declare competing interests.

Figures

Extended Data Figure 1. 6-Hydroxydopamine (6OHDA) treatment ablates sympathetic nerve function and reduces ear swelling, but does not ameliorate the inflammatory response to IMQ treatment

a, Experimental protocol: Mice were injected intraperitoneally with 6OHDA, resulting in a reversible chemical sympathectomy lasting ~two weeks. After a rest period of 3 days animals were challenged topically on the ear with IMQ. b, Representative section of splenic white pulp showing B cells (B220, white), T cells (CD3, red), and tyrosine hydroxylase+ (TH, green) nerve fibers in vehicle (ascorbic acid) treated and sympathectomized (6OHDA) mice. c–i. Analysis of the inflammatory response in ears of vehicle (ascorbic acid) treated and sympathectomized (6OHDA) mice after daily topical IMQ challenge: (c) timecourse of change in ear thickness of IMQ treated ear relative to the contralateral ear (n=10; *, P < 0.05; **, P < 0.01) and (d) total number of infiltrating monocytes and (e) neutrophils, and the amount of (f) IL-17A, (g) IL-17F, (h) IL-22 and (i) IL-23-p40 in protein extracts of IMQ exposed ears at day 3 (*, P < 0.05; n=5).

Extended Data Figure 2. Resiniferatoxin (RTX) treatment diminishes noxious heat sensation and decreases the expression of nociceptor markers on dorsal root ganglia (DRGs)

a, Schematic protocol of nociceptor ablation and induction of psoriasisform skin inflammation. RTX was injected subcutaneously into the back in three escalating doses (30 μg/kg, 70 μg/kg and 100 μg/kg) on consecutive days and mice were allowed to rest for at least 4 weeks before IMQ treatment. b, Denervation was confirmed by immersing the tail of mice into a temperature controlled water bath maintained at 52°C and the latency to the first tail movement to avoid water was measured (n=6). c, Total RNA was isolated from dorsal root ganglia (DRG; level C1-C7) of vehicle (DMSO) and RTX-treated mice and the levels of trvp1, scn10a (NaV1.8), tac1 (Substance P), trpm8 and trpa1 mRNA relative to gapdh were determined (n=3).

Extended Data Figure 3. Gating strategy for T cell subsets and myeloid cells from digested ear skin

a, The ear skin of mice challenged for 3 days with IMQ was digested as described in methods and, after doublet exclusion and gating on defined FSC-A, SSC-A parameters, infiltrating myeloid cells were gated as CD45+ I-Ab (Class-II)−, CD11b+ CD11c−, and then subdivided into inflammatory monocytes and neutrophils based on Ly6C and Ly6G staining. b, The ear skin of naïve mice was digested as described in methods and, after doublet exclusion and gating on defined FSC-A, SSC-A parameters, cutaneous T cells were gated on CD45+, Thy1+, and then divided into subsets based on staining for δ-TCR and β-TCR.

Extended Data Figure 4. RTX treatment reduces the immune cell infiltrate upon IMQ treatment in the skin but does not affect reservoirs of inflammatory monocytes and neutrophils at steady-state

a, The ear skin of vehicle (DMSO) or sensory denervated (RTX) mice was treated with topical IMQ cream daily and the total numbers of CD45+ cells were determined on day 3 as explained in methods (n=10). b, Representative histological sections of untreated and IMQ treated ears at day 3 stained by H&E (20x) (n=5 per condition). c,d, Total inflammatory monocytes (CD45+, CD11b+, Ly6Chigh) and neutrophils (CD45+, CD11b+, Ly6Ghigh) were determined by flow cytometry (n=5–10 mice per time point). Two-way ANOVA was run to compare total numbers of inflammatory monocytes and neutrophils between DMSO and RTX conditions over days 3–6 (****, P< 0.0001). One-way ANOVA was run to compare total inflammatory monocytes and neutrophil numbers over days 3–6 within DMSO or RTX conditions (**, P< 0.003). e, Bone marrow was isolated from WT and RTX mice from one femur and the frequency of inflammatory monocytes (CD45+, CD11b+, Ly6Chigh, Ly6G−) and neutrophils (CD45+, CD11b+, Ly6Ghigh, Ly6Cmid) relative to CD45+ cells was determined by flow cytometry (n=5). f, Spleens from WT and RTX mice were processed for flow cytometry and the frequency of inflammatory monocytes (CD45+, CD11b+, Ly6Chigh, Ly6G−) and neutrophils (CD45+, CD11b+, Ly6Ghigh, Ly6Cmid) relative to CD45+ cells was determined (n=5).

Extended Data Figure 5. Leukocyte rolling fractions in skin venules of control and RTX-treated mice analyzed by intravital microscopy

Combined results are shown for 26 venules from 5 control mice and for 20 venules from 4 RTX-treated mice. Data are expressed as mean ± SEM of four experiments.

Extended Data Figure 6. Dermal γδ T cells represent a major source of IL-17F and IL-22 in skin during IMQ challenge and already express IL-23R+ at steady state

a, WT mice were challenged with IMQ and the total numbers of IL17F+ dermal γδ T cells and αβ T cells at 3 days (n=15) or 6 days (n=10) were quantified. b, Representative flow plots related to (Figure 2g) of gating for IL-22+ cells within dermal γδ T cells after 6 days of IMQ treatment. c, Ears of DMSO or RTX treated mice were exposed for 6 days to IMQ and the frequency of IL-17F+ and IL-22+ cells within αβ T cells was determined (n=5).d, Auricular lymph node (aLN) cells from IL-23RGFP/+ mice were analyzed by flow cytometry for expression of IL-23R-GFP+ cells within the γδ T cells and αβ T cell compartment at steady state (representative FACS plot from 8 mice analyzed). e, The ear skin from IL-23RGFP/+ mice was digested and analyzed by flow cytometry and the distribution of T cells subsets within IL-23R-GFP+ and IL-23R-GFP− fractions of Thy1+ cells determined (representative FACS plot from 8 mice analyzed).

Extended Data Figure 7. TRPV1+ nociceptors regulate the expression of il12b and il23a upon IMQ challenge, the inflammatory response to DNFB application, and IL-23 injection can bypass their contribution to activate γδ T cells

a–c, After 3 days of IMQ challenge, ears were harvested and processed for total RNA isolation and (a)il12b(b)il23a and (c)il12a mRNA levels were analyzed by qPCR (n=5). d, DNFB (0.5% in acetone) was applied topically to DMSO and RTX mice. Time course of change in ear thickness of IMQ treated ear relative to the contralateral ear is represented (n=10). Two-way ANOVA was run to compare ear swelling under DMSO and RTX conditions over time (****, P < 0.0001). e, Representative FACS plots from ears harvested after 24h of DNFB application. f, IL-23R−/− mice were treated with RTX and then compared to WT and IL-23RGFP/GFP littermate controls during IMQ treatment. Ear thickness was calculated relative to the contralateral ear (n=5). g, After two IL-23 injections into the ear skin of WT and IL-23RGFP/GFP mice, the frequency of IL-17F+ cells within dermal γδ T cells was determined by flow cytometry (n=5). h, IL-23 was injected twice into the ear skin of Vehicle- and RTX-treated mice and the total numbers of IL17A+ or IL-17F+ dermal γδ T cells per ear were determined by flow cytometry (n=5).

Extended Data Figure 8. Selective depletion of migratory and skin resident myeloid cell subsets in ear skin and gating strategy used for sorting to isolate RNA from MHC-II+ cells in skin

a, WT mice were treated with anti-Gr1 (clone RB6-8C5 to deplete neutrophils and inflammatory monocytes) or matched isotype control, challenged with IMQ for 3 days and skin was digested to quantify the total numbers of inflammatory monocytes and neutrophils per ear. Shown are representative plots pre-gated on CD45+ cells and quantification of cell numbers from n=3 mice. b, DTX treatment resulted in depletion of both subsets of dermal DCs (DDCs) as well as Langerhans cells (LCs) but not macrophages. Cells were gated as shown in (Extended Data Fig. 8c) and normalized to levels in WT mice based on the frequency within the CD45+ population from n=4 mice. c, Ear skin from naïve mice was digested and analyzed by flow cytometry for the indicated subsets. Shown is a representative plot pre-gated on CD45+ Class II+ cells from which further subsets were divided based on CD11b and CD11c expression and then F4/80 and CD103 as indicated. d, Total RNA from sorted cells was isolated and qPCR for il12a relative to gapdh was performed from naïve and IMQ treated ears after 6 hours from n=20 pooled mice.

Extended Data Figure 9. Dermal DCs (DDCs) are found in close apposition to NaV1.8+ nociceptors in skin, NaV-DTA mice express reduced levels of key nociceptor markers, yet nociceptor deletion does not grossly affect the peripheral neural network in skin

a, Representative confocal micrographs of CD11c-YFP mice stained for β3-tubulin, Lyve-1 (collecting lymphatics), and CD31 (blood and lymphatic endothelial cells). b, 3D quantification of DDC proximity to peripheral nerves in naïve and 6 hours post-IMQ treatment binned into contact (<0 um), proximal (0–7 um) and distal (>7 um) fractions as explained in the methods (n of DCs = 200). c, Total RNA from dorsal root ganglia (DRGs) (C1–C4) of littermate control and NaV1.8-DTA mice was isolated and levels of mRNA for trpv1 (TRPV1), scn10a (NaV1.8), tac1 (Substance P) and trpa1 (TRPA1) were determined relative to gapdh. This demonstrates the efficacy of the NaV1.8-DTA system and combined with the original reference characterizing the pain phenotype of these mice illustrates that a subset of peptidergic TRPV1+ nerve fibers is spared. d, Representative confocal micrograph of whole mount ear skin of Vehicle- and RTX-treated mice showing preserved nerve scaffold and e, representative confocal micrographs of whole mount ear skin of Control and NaV1.8-DTA mice showing preserved nerve scaffold. While DRGs showed a loss of the hallmark ion channels of these nerve subsets (Extended Data Fig. 1c and Extended Data Fig. 9c), surprisingly we still observed that RTX mice and NaV1.8-DTA mice maintain a meshwork of nerves in the skin.

Extended Data Figure 10

Summary Diagram.

Figure 1. TRPV1+ nociceptor ablation attenuates skin inflammation and draining lymph node hypertrophy in the IMQ model

a, Representative whole-mount confocal micrograph of normal ear skin from NaV1.8-TdT reporter mice (NaV1.8+ nociceptors, red) stained for β3-tubulin (peripheral nerves, blue) and tyrosine hydroxylase (TH, sympathetic nerves, green). b–f, The ear skin of vehicle treated controls (DMSO) or TRPV1+ nociceptor ablated (RTX) mice was treated with topical IMQ cream daily. (b) Ear thickness was measured relative to the contralateral ear at indicated time points (n=10–15 mice per time point; *, P < 0.02). (c) Representative histological sections of IMQ treated ears at day 6 stained by H&E. (d) Total inflammatory monocytes (n=10) and (e) total neutrophils in skin at day 3 (n=10; **, P < 0.005). (f) Total cell number in auricular lymph nodes at day 3 (n= 20; *, P = 0.01; ***, P < 0.001). g–j, IMQ was applied daily to (g,h) WT (n=10) and LTα −/− mice (n=6) or (i,j) vehicle treated (n=10) and FTY720 treated mice (n=10) and (g,i) ear swelling was measured at indicated time points. (h,j) The percentage of CD45+ leukocytes was determined in ear skin digests on day 6.

Figure 2. TRPV1+ nociceptors control IL-17F and IL-22 production by IL-23R+ dermal γδ T cells

a–c, After 3 days of IMQ challenge, ears from vehicle- (DMSO; n=5) or RTX-treated mice (n=5) were harvested to perform ELISA for (a) IL-17A, (b) IL-17F, and (c) IL-22. (**, P < 0.01; ***, P < 0.001). d–g, Flow cytometry on digested ear skin was performed on day 6 of IMQ challenge. (d) Relative frequency of IL-17F+ or IL-22+ dermal γδ T cells and αβ T cells in IMQ treated control mice (n=15). (e) Representative FACS plots of IL-17F staining in dermal γδ T cells from vehicle (DMSO) and RTX treated mice. Quantification of frequency of (f) IL-17F+ and (g) IL-22+ amongst dermal γδ T cells in DMSO and RTX mice (n=5/group; *, P < 0.05; **, P = 0.01). h, Representative FACS plots of normal ear skin from an IL-23RGFP/+ mouse and i, quantification of frequency of IL-23R-GFP+ cells among skin resident Thy1+ T cell subsets (n=8).

Figure 3. Dermal DC-derived IL-23 is critical to drive psoriasiform skin inflammation and acts downstream of RTX sensitive nociceptors

a, After 3 days of IMQ challenge of vehicle (DMSO) or RTX treated mice, ears were harvested and total protein was prepared to quantify IL-23p40 by ELISA (n=5/experiment;***, P < 0.001). b, Ears of WT or IL-23RGFP/GFP mice (n=5/group) were treated daily with IMQ and ear thickness was measured relative to the contralateral ear at indicated time points (***, P < 0.001). c–f After 3 days of IMQ challenge in WT (n=5) or IL-23RGFP/GFP mice (n=4) total protein was prepared from ear skin and (c) IL-17F and (d) IL-22 were quantified by ELISA (*, P < 0.05; ***, P < 0.001). Cell suspensions from exposed ears stained for (e) inflammatory monocytes and (f) neutrophils to assess total numbers (n=5/experiment; *, P < 0.05; **, P < 0.01). g–i, IL-23 was injected intradermally every other day into ears of DMSO or RTX treated mice (n=8/group). (g) Ear thickness was measured as indicated. After 3 days, (h) IL-17A and (i) IL-17F producing Thy1+ cells per ear were quantified by FACS (n=5). j,k, Mice were treated with anti-Gr1 to deplete neutrophils and inflammatory monocytes or isotype-matched control mAb and challenged with IMQ for 3 days. Ear skin protein lysates were analyzed for (j) IL-23p40 and (k) IL-17F by ELISA (n=5). l, CD11c-DTR mice were treated with diphtheria toxin (DTX) or PBS 12 h prior to IMQ challenge. After 6h, ears were harvested and processed for total RNA isolation and il23a mRNA levels analyzed by qPCR (n=4; ***, P = 0.001).

Figure 4. Dermal DCs (DDCs) are closely associated with cutaneous nerves and depend on NaV1.8+ nociceptors for IMQ induced IL-23 production

a, Mice were challenged with IMQ (n=20 pooled mice/condition) and after 6h, myeloid cell populations comprising dermal macrophages and two subsets of DDCs were FACS sorted (Extended Data Fig. 8c) from cell suspensions to measure il23a mRNA by qPCR. b, Representative confocal micrographs of ear skin whole mounts from CD11c-YFP mice stained for β3-tubulin (peripheral nerves, red) and Lyve-1 (collecting lymphatics, blue) or CD31 (blood and lymphatic endothelial cells, blue). Original magnification was 200X. c, Close-up confocal micrograph of a CD11c-YFP cell in contact with a nerve (see also Suppl. Videos 1&2). d, Quantification of 3D DDC proximity to peripheral nerves, lymphatics, and blood vessels in normal ear skin. The frequency of DDCs (n=330) in contact, proximal (0–7μm) and distal (>7μm) to nerve fibers was determined as described in Methods and a Chi-square test showed bias of DDC to nerves relative to lymphatics and blood vessels (***P< 0.0001).e–h, Ears of NaV1.8-DTA or control littermates were treated daily with IMQ. Total protein was prepared from ear skin after 3 days and (e) IL-17A, (f) IL-17F, (g) IL-22 and (h) IL-23p40 were quantified by ELISA (n=4 ears/condition; *, P < 0.05; **, P < 0.01).